Santa’s smoky chimney syndrome 🧱🧱🔥

Author

Paula Štancl, PhD

Santa Claus Is Coughing…

While delivering gifts last Christmas, Santa spent hours squeezing down chimneys,
exposed to soot, smoke, and cold winter air.

This year, the merry old guy has developed a persistent cough, occasional shortness of breath,
and the elves are worried. Could these symptoms — in combination with years of chimney exposure —
signal something more serious lurking behind all that festive cheer? 🎄

Before Santa panics and trades his sleigh for a hospital bed,
he’s turning to you — the finest machine-learning researchers in the North Pole! 🕵️🤖

🎯 Your Mission

Develop a predictive model that can distinguish early-stage vs late-stage carcinoma based on real clinical and gene expression data.

Because Santa’s medical scans and biopsy results are strictly classified (sealed in a candy-cane-encrypted vault 🍬🔐),
you’ll only unlock access once your model shows strong performance on the provided training data.

Your task:

🧬 Identify the key clinical features linked to cancer progression
📈 Build and validate a model to classify carcinoma stage
🔍 Determine whether Santa may require further clinical examination

With your help, Santa will breathe freely again (and continue expertly sliding into homes worldwide 😌).
Let’s make sure this holiday season remains full of joy — not cough drops! 🌬️🎁✨

Let’s Get to Work! 🔬🎄

Download the dataset

🔍 Exploratory Data Analysis (EDA)

Begin by thoroughly investigating your dataset and its variables.
Look for patterns, differences between groups, and potential predictors of cancer stage.

Your task:
Formulate up to 10 meaningful scientific questions and explore them with appropriate visualizations or summary statistics.

Such as Santa asks himself are girls more happier when receiving present wrapped in pink with a glittery bow?

🤖 Model Development: Predicting Carcinoma Stage

Build at least three-four classification models using your training data.

Then:

Compare their performance
Select the best model to move forward
Which medical features are most important for predicting cancer stage?

Elf Paula’s Approval Checkpoint 🧝

Before Santa’s confidential data is unlocked,
you must submit your best model to the Elf Review Committee™ for approval.

Once the elves confirm that your model meets North Pole regulatory standards (NP-FDA),
they will provide:

Santa’s private biomedical data in a independent test set

Your final task:
🎯 Determine Santa’s predicted cancer stage
📈 Assess how well your model generalizes to unseen cases 🍬

Let’s hope that Santa ends up on the healthier, happier side of your classification results —
we need him fit for sleigh duty!

--- title: "Santa’s smoky chimney syndrome 🧱🧱🔥" author: "Paula Štancl, PhD" format: html: self-contained: true toc: true toc-depth: 5 code-fold: false fig-align: center df-print: paged code-summary: "Show code" code-line-numbers: false code-tools: true execute: echo: true warning: false message: false --- # Santa Claus Is Coughing... While delivering gifts last Christmas, Santa spent **hours squeezing down chimneys**,\ exposed to **soot, smoke, and cold winter air**. This year, the merry old guy has developed a **persistent cough**, occasional **shortness of breath**,\ and the elves are *worried*. Could these symptoms --- in combination with years of chimney exposure ---\ signal something more serious lurking behind all that festive cheer? 🎄 Before Santa panics and trades his sleigh for a hospital bed,\ he's turning to **you** --- the finest machine-learning researchers in the North Pole! 🕵️🤖 ![Oh noooo!](img_santa/santa_chimney.png) ------------------------------------------------------------------------ ## 🎯 Your Mission Develop a **predictive model** that can distinguish **early-stage vs late-stage carcinoma** based on real clinical and gene expression data. Because Santa's medical scans and biopsy results are **strictly classified** (sealed in a candy-cane-encrypted vault 🍬🔐),\ you'll only unlock access once your model shows strong performance on the provided training data. Your task: - 🧬 Identify the **key clinical features** linked to cancer progression\ - 📈 Build and validate a model to classify **carcinoma stage**\ - 🔍 Determine whether Santa may require **further clinical examination** ------------------------------------------------------------------------ With your help, Santa will breathe freely again (and continue expertly sliding into homes worldwide 😌).\ Let's make sure this holiday season remains full of joy --- **not cough drops**! 🌬️🎁✨ ------------------------------------------------------------------------ # Let's Get to Work! 🔬🎄 [Download the dataset](projects/Lung_cancer_subset.csv.gz) ### 🔍 Exploratory Data Analysis (EDA) Begin by thoroughly investigating your dataset and its variables.\ Look for patterns, differences between groups, and potential predictors of cancer stage. Your task:\ Formulate up to **10 meaningful scientific questions** and explore them with appropriate visualizations or summary statistics. Such as Santa asks himself are girls more happier when receiving present wrapped in pink with a glittery bow? ### 🤖 Model Development: Predicting Carcinoma Stage Build **at least three-four** classification models using your training data.\ Then: 1. **Compare their performance** 2. **Select the best model** to move forward 3. Which medical **features** are most important for predicting cancer stage? ## Elf Paula's Approval Checkpoint 🧝 Before Santa's confidential data is unlocked,\ you must **submit your best model to the Elf Review Committee™** for approval. Once the elves confirm that your model meets North Pole regulatory standards (NP-FDA),\ they will provide: - Santa's **private** biomedical data in a **independent test set** Your final task:\ 🎯 Determine Santa's predicted cancer stage\ 📈 Assess how well your model generalizes to unseen cases 🍬 Let's hope that Santa ends up on the **healthier, happier** side of your classification results ---\ we need him fit for sleigh duty!